论文部分内容阅读
设置了转子不平衡、联轴器不对中、风机基座松动、转子径向摩擦和轴承内圈磨损等5种故障,针对风机的振动问题进行了实验研究.在简要分析ICA理论及其算法的基础上,提出应用基于高斯矩的Fast ICA算法对模拟信号及实测离心式风机振动信号进行去噪处理,并分别与EMD及db8小波的滤波去噪效果进行定量比较.结果表明:Fast ICA方法与EMD方法和小波方法一样,能有效地处理短时瞬态及含宽带噪声的信号,但Fast ICA方法不受去噪阈值的影响,也不需要选择小波基函数,更具有通用性和稳定性.
Five kinds of faults such as rotor imbalance, misalignment of coupling, loosening of fan base, radial friction of rotor and wear of bearing inner ring are set up, and the vibration of fan is experimentally studied.A brief analysis of ICA theory and its algorithm Based on the fast ICA algorithm based on Gaussian moment, the noise signals of the analog signals and measured centrifugal fans are denoised and compared with those of the EMD and db8 wavelet respectively. The results show that the fast ICA method and Like the wavelet method, the EMD method can effectively deal with short-term transient and broadband noise-containing signals. However, the Fast ICA method is not affected by the denoising threshold and does not need to select the wavelet basis function. It is more universal and stable.